| Research Abstract |
Understanding complicated physical phenomena is a major theme of computational physics. Especially, turbulent flows are one of the most interesting subjects, and have been studied energetically by many researchers and engineers. In the background of the situation is the finding that turbulence, although of a random nature, has a coherent structure, and can be estimated deterministically by a Navier-Stokes equation. Approaches based on computational fluid dynamics have already increased the understanding of turbulent flows. In particular, a method called DNS plays a greater role owing to its high accuracy. Since it is based only on physical laws without a turbulence model, DNS is expected to be universally applicable to any flow field. DNS such as the spectral method, however, is not applicable to complex boundary flows while the applicability is required from an engineering viewpoint. DNS methods using a finite difierence method have recently been increasing due to their expasibility t
… More
o more comptex systems. Analyses have, however, still been made only of simplified turbulence fields including channel flows. Few reports are available on general flow fields with a curvilinear boundary(Krettenauer et al.,1992, De Angelis et al.,1997, Mito et al., 1998,Ohta et al.,1998).
Technological development for flow visualization, in the meantime, has revealed the existence of a coherent structure of relatively long life in turbulent flows of a complex time-space structure(Kline et al.,1967), and has also been demonstrating that the coherent structure is closely related to the transport of turbulent flows involving the generation, mixing and diffusion of turbulence. The coherent structure forms a slender cylindrical high-vorticity area surrounded by spiral flows, and is called cylindrical spiral vortex(low-pressure vortex). It plays a important role in generating streaks and Reynolds shear stress near the wall, and is thus regarded as a basic structure that expresses the turbulence structure. Much, however, needs to be clarified including the relationship to the secondary currents of Prandtl's second kind. Expectations are, therefore, placed on future development in particle-image velocimetry(PIV). Understanding the turbulence structure requires not only the data on the time-average and instantaneous fields but also information on ever-changing time development of the flow field. At present, therefore, numerical analyses including DNS are considered highly effective methods. Less
|
